Photogrammetry is similar to the 3D scanning process with photographs. It probably seems too fantastic to be true. Just imagine, take a few pictures of the object, send them to the algorithm and get a textured 3D model.

After presenting the model, the second question asked is the question of accuracy. What is the accuracy of 3D scanning from a photo? Answer: submillimeter range. The average deviation can reach 0.78 mm, that is, less than one millimeter compared to a scanned 3D model made using a laser scanner.
Just like in the market for laser scanners, in photogrammetry there are many different software options for checking. They range from proprietary solutions with restricted access to open and free.

In order to obtain materials for loading into a program that creates 3D models based on photo data, it is necessary to photograph the desired object. Photogrammetry can be used absolutely for any three-dimensional models, and it is based on its type that you will need to choose a shooting method. If this is a separate object (for example, some kind of archaeological find that requires reconstruction), its shooting is done manually. The same is true for much more, but if you plan to visualize the exterior in 3D max or create a three-dimensional landscape, it will be more efficient to shoot using a quadcopter equipped with a camera.

The method of photogrammetry is based on loading the resulting photos into programs designed to create three-dimensional models based on these images, and the number and features of the pictures will depend on which program is planned to be used. Based on the received photos uploaded to the program, a three-dimensional model of the object is created.


The undoubted advantage of photogrammetry is the saving of time in comparison with conventional modeling, which is especially seen when working with large-scale and complex geometry scenes or objects, for example, large buildings or whole landscape scenes. Thanks to the program, which analyzes the obtained images and creates a three-dimensional model, you can get rid of creating the same model from scratch and, having received a polygonal model, proceed to edit it.

One of the difficulties that you may encounter when applying this method is how to properly take a picture of the subject. To form a model, it is necessary to shoot an object in such a way as to provide approximately 60% overlap between adjacent frames; Thus, we take several dozen photos in one circle, and there may be several such circles. In addition, it is recommended to avoid unnecessary objects that fall into the frame, so as not to spend a lot of time on cleaning textures.